Wolfram|Alpha questions

I think Doug’s question about well-formed questions deserves some more thought. Wolfram think they’ve cracked the issue of the questions we’ve asked with their natural language processing, but can they?

What will it do with questions deliberately constructed to be difficult, either because of the amount of computational power required, or because they are self-referential? How will it handle paradox such along the lines of Smullyan’s “what is the name of this book?” Godel’s incompleteness theorem surely limits the validity, in the limit, of the answers it can produce. The other way to test it would be with questions such as “what is the nth prime?” where n is large.

I’m fascinated by what it will look like. I’d love it if the black box had glass sides – it will be interesting to see what they publish (eg: statistics for requests, requests actioned, solutions, etc.

Will there be an faq section?

Will there be a Web2.0 element? To my mind, a database of answers and a system for users to score and/or tag the answers would be valuble. This could either be used to iterate the algorithm to improve on near misses, or to suggest similar questions and answers along with the specific “custom” solution.

PS: has anyone noticed that if you write the initials in Greek you get omega|alpha? With the pipe standing for from, or given, (as it does in some set theory notations) it then becomes omega given alpha, which looks suspiciously close to becoming alpha and omega.

Technorati Tags: , , , ,

This entry was posted in Wolfram|Alpha and tagged , , , , . Bookmark the permalink. Post a comment or leave a trackback: Trackback URL.

Post a Comment

Your email is never published nor shared. Required fields are marked *

*
*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>